Method
To test my hypotheses I conducted exploratory factor analysis on LAPOP survey data taken in Costa Rica in 2004 and 2008. This analytical technique, based on bivariate correlations among a large number of variables, seeks patterns or dimensions among the larger number of variables. Results are reported in the form of a factor structure of dimensions identified, plus a factor loading (essentially a correlation coefficient of each variable on each factor). For a table of the factor loadings for the 2004 and 2008 LAPOP Costa Rica surveys, see Appendix I. Survey questions are considered to be associated with a particular dimension (factor) if their loadings have an absolute value greater than 0.5 with that dimension.
The factor analysis used an SPSS Oblimin Rotation method with Kaiser Normalization to account for and measure correlation among the dimensions found. This is important because rather than these being uncorrelated dimensions, I expected some correlation to be found among them. In The Legitimacy Puzzle, Booth and Seligson used this method as well as one in which the structure of legitimacy was modeled and then that model was tested via confirmatory factor analysis (2009). In my principal component analysis, I instead look for dimensionality among the survey factors without suggesting a model, allowing for insight into what considerations may be needed to adjust the model to a longitudinal study. After determining the factor composition of each dimension for both surveys, I calculate the descriptive evaluation level of each legitimacy dimension by averaging the means of the survey questions that compose each dimension. For a table of dimension evaluation level for both surveys, see Appendix II.